💬│general

[vale lark]

Bruh, i don't have that

[sacred heron]

Sorry, don't mean to interrupt. Could someone help me with this? I got a notification saying I could get 12 months of Perplexity Pro for free through the Samsung app store but I can't seem to properly access the offer.

[short narwhal]

Any DNS filter or VPN on?

[vale lark]

DNS filter yes, NextDNS.

I try to disable that

[agile charm]

Same here as a pro sub.

[fast meteor]

hmmm weird. can you also try it on incognito browser?

[short narwhal]

Had the same issue with hagezi Ultimate list with Perplexity in the past

[fast meteor]

elegant

[serene sand]

#💬│general message

[agile charm]

Same result.

[fast meteor]

which ai model did you use?

[languid star]

youtube and google

[vale lark]

Yeah, is the DNS filter

"hmmm weird. can you also try it on incognito browser?"

Thinking it was some kind of ultra-persistent cache or cookies that couldn't be deleted even with Incognito Mode, I ended up installing another browser

[short narwhal]

Christopher Lennertz

[blazing comet]

Comet browser

[fast meteor]

interesting

[sick crypt]

Hmm

[brazen plinth]

hi

everyone

[vale lark]

I disable the HaGaZi rule and it work

So much false positive

[blazing comet]

https://fixupx.com/aravsrinivas/status/1953922333993054281?s=46

Aravind Srinivas (@AravSrinivas)

For Max subscribers on Comet: give the Max Assistant a shot. Will auto-route to the smartest models and be a lot more reliable for longer horizon or more complex parallel agent tasks.

Quoting Seth Saler (@sethsaler)
︀
"Enable Max Assistant" in @PerplexityComet (1) assistant window.
︀︀
︀︀Ability to add a screenshot from the webpage (2) (new?)

**💬 12 🔁 4 ❤️ 56 👁️ 7.1K **

FixupX · Aug 8, 2025, 8:52 PM

🥀

[short narwhal]

I think hagazi pro and under that work just fine with PPLX. Ultimate and higher blocks some things

[brazen plinth]

what u all discussing >>?

[vale lark]

Well, it had to be for Max users.

[fast meteor]

how do you all know this?

[vale lark]

But i'm using de hgz pro, strange

[brazen plinth]

???

[short narwhal]

🤔 I might be remembering wrong too

[timber niche]

@fast meteor your hat is gone 🙁

[short narwhal]

Nuh uh

[vale lark]

I will try to clear the DNS cache with HaGeZi Pro to see if it happens again.

[dry kestrel]

Blind

[timber niche]

Weird

It shows up

Only when i open it

[short narwhal]

Normal works just fine at least

[agile charm]

At perplexity ai, or in Comet?

[vale lark]

@short narwhal Yeah, work well with hgz pro

[timber niche]

I think

Maybe you changed it

[vale lark]

Everything is broken with ultimate

[dry kestrel]

It's pretty well known

[timber niche]

Then changed back

[short narwhal]

lol yeah

[timber niche]

It's fine now

Nice hat

[brazen plinth]

@timber niche hi

[dry kestrel]

samsung

[brazen plinth]

@short narwhal hi

[vale lark]

Is a DNS list

[short narwhal]

@brazen plinth hi

[timber niche]

Hi

[fast meteor]

i changed the global avatar, but kept the per server one

[timber niche]

Oh wow

[dry kestrel]

Dammit I need nitro now

[timber niche]

Betrayal

[short narwhal]

Me tooo

[fast meteor]

now face the darkness

[short narwhal]

Let me be free of this bucket

[timber niche]

@dry kestrel you have it tho

Nitro

[dry kestrel]

I have the classic one

[short narwhal]

aabb has gone emo

[agile charm]

Ohhhh...ok, thanks!

[timber niche]

Oh just decoration

[dry kestrel]

It's literally just for higher res streaming and emojis

[timber niche]

Good enough

[dry kestrel]

NO

I might actually get nitro just to change the avatar

[timber niche]

Atp

Just forget about the hat

[meager rover]

Does the video gen also make sounds?

[timber niche]

Don't waste money ig

[meager rover]

Because I don’t hear shit

[short narwhal]

Don't think so

[timber niche]

Download locally

And play

Veo sometimes has me have to do that

[meager rover]

😡

[fast meteor]

https://s3.amazonaws.com/lists.disconnect.me/simple_ad.txt
https://s3.amazonaws.com/lists.disconnect.me/simple_tracking.txt
https://raw.githubusercontent.com/StevenBlack/hosts/master/alternates/fakenews/hosts
https://big.oisd.nl/
https://shreshtait.com/newly-registered-domains/nrd-1m

what dns block lists you all subscribed?

[short narwhal]

If PPLX is using Veo 3 fast that won't work

[meager rover]

time to protest

[short narwhal]

also... :(

[timber niche]

Click video

Rigth click

Download

Worked for me

[dry kestrel]

There is Veo3 fast audio and no audio

[timber niche]

Huh

[dry kestrel]

over the API

[short narwhal]

aaaa

[short narwhal]

err

[meager rover]

and to clarify, rate limits are 3 right?

[dry kestrel]

I feel the need to buy nitro now

[dry kestrel]

Yes

per month

[meager rover]

WHAT

[short narwhal]

I'll change my pfp if yours is only in this server

[timber niche]

Fair enough

[dry kestrel]

Thats... Not fair!!

[hollow vessel]

Now I gotta wait to get gpt 5 on my teams account lol

[timber niche]

I still have my hat

[short narwhal]

The deal was if you get bullied = me free

[dry kestrel]

noo

[timber niche]

Sorry i don't speak mac

[dry kestrel]

Luckily I have this avatar deco

[vale lark]

I saw the video generation, and based on the quality and the “film grain” effect, I think it's Wan2.2.

Excellent addition, because if we're talking about Veo3, their bills are sky-high.

[dry kestrel]

veo3 fast is not that bad, esp without audio

[short narwhal]

oh well I'm saving it as html

best video format

[timber niche]

Remember the grass video you shared

[fast meteor]

i can change my global avatar back if that keeps your bucket on

[timber niche]

I managed to download it

And share here

[dry kestrel]

at least your hat looks fine

I've got this cartooney propeller hat

[vale lark]

Can veo3 be used without audio? Because the video I generated doesn't have any, and based on the quality, everything points to Wan2.2 at 480p

[short narwhal]

that's heat wdym

[dry kestrel]

Yes

[timber niche]

@dry kestrel it's not even visible with rhe decoration

So unfair

[meager rover]

😱

[short narwhal]

imagine wearing double hats and wanting to get rid of the cool hat

[dry kestrel]

Hehe

[fast meteor]

this fits with the cockatiel besides you hide the propeller part already

[dry kestrel]

It's going to expire :(

[short narwhal]

Are you gonna hit your 3 vidoes per month limit soon

[meager rover]

I already did

[dry kestrel]

Once it does I might get nitro just to keep it

[meager rover]

I was trynna get the satisfying glass apple cut things

[vale lark]

There is no way Perplexity would offer this to users, as they use it eight times and are already losing money on LLM, Embedding, and other APIs.

[meager rover]

I’ll just use sora 😡

[dry kestrel]

*13 times

[short narwhal]

Je vois

[dry kestrel]

hence the 3 a month limit

[fast meteor]

@short narwhal pls keep the bucket

[vale lark]

That's assuming your subscription was only to generate those 13 videos.

[harsh verge]

oh, tu parles français !

[short narwhal]

oki

[dry kestrel]

You have a monthly limit of 3 videos...

[short narwhal]

🙅

[vale lark]

Oh! Make sense

[frosty stone]

3 a month on pro?

[meager rover]

Yup

[dry kestrel]

日本語?

[dry kestrel]

yes

[vale lark]

I already spent the first one uselessly, terrible, let's cry.

[meager rover]

I’m good

[short narwhal]

はい！私が勉強です

No clue if that is right

[dry kestrel]

勉強?

so kinda barely

[fast meteor]

you barely speak?

[meager rover]

how long has your journey been for learning Japanese @short narwhal

[short narwhal]

1 month and 8-9 days

[meager rover]

Rad

Good luck

[short narwhal]

ありがとうございます thank youu

[minor shard]

https://tenor.com/view/im-proud-of-you-dan-levy-david-david-rose-schitts-creek-gif-20773745

[short narwhal]

wdym

[agile charm]

I'm totally missing something I'm trying to follow in this conversation. I'm a perplexity pro subscriber - do I have access to a video gen model or not? I don't see one in the model lists I can see...

[meager rover]

what made you wanna begin learning it tho

[dry kestrel]

アィガト - this might be gibberish

[short narwhal]

whaa

[meager rover]

I mean I know the why not factor but like 😂

[short narwhal]

yeah gibberish

[dry kestrel]

Yes, you get 3 times a month

[minor shard]

I just ask ChatGPT to speak Japaneese for me

😄

[dry kestrel]

日本語

ok thats the only one i know how to type

[agile charm]

What's the model mane?

[dry kestrel]

likely veo3fast

[meager rover]

not bad

[fast meteor]

勉強=study

[short narwhal]

I went to see an artist but then they started going on about stuff in Japanese during their MC (as they are a Japanese artist) and I had to clue what they were saying, and after the concert outside I had the incredible idea to start learning it

[minor shard]

https://tenor.com/view/spiderman-asmr-gif-9370173392474378875

[dry kestrel]

ok yeah this language makes no sense

しぶ

[meager rover]

Damn crazy origin story, good luck keep it up 🫶

look at this banger I made

goat simulator

[short narwhal]

woah

[meager rover]

😼

[silent ocean]

MANUS agent

[meager rover]

yeah manus is crazy

[silent ocean]

The results are rather underwhelming though

[meager rover]

Hm.

Well I good really well made reports when I asked it to

[silent ocean]

Outdated info

This was the easiest task ice given it

All is searchable in seconds

Simple comparison

Still failed on specs

Prices

Perplexity is still the most reliable researching tool

Maybe my use case is rather niche then

Poland, why do you ask?

???

[meager rover]

um

bruh I just work here

https://tenor.com/view/i-just-work-here-dont-blame-me-im-innocent-coward-scared-gif-15728242

[agile charm]

Apologies for being repetitive, but Veo simply is not among the models I see whether I'm using the web site or desktop app. Any suggestions besides trying to get a response from the recently magically disappeared support team? Max sub required or something?

[meager rover]

you dont see it.

you just say yo make a video for me then it'll make it

[silent ocean]

Can you please not spread religious hatred? @tame current

[dry kestrel]

Given the quality, its the most likely model that they use.

[meager rover]

under the hood

[tranquil orchid]

7. Avoid political and religious discussions.

To keep the space welcoming for everyone, avoid political and religious topics and debates.

[silent ocean]

5 videos a month is rather dissapointing

But those 5 videos are good quality

[dry kestrel]

It's 3 btw

[silent ocean]

Oh

Maybe I got lucky

[sweet zealot]

damn i tought PPLX is the best option...

they don't even have agent mode.

[silent ocean]

They do

[agile charm]

Oh. I feel like the idiot I am. That's what I get for not reading instructions like a non-moron would do.

[silent ocean]

On comet

I hope we can get an agent mode like chatgpt or Manus in PPLX

[frosty stone]

As a max user id like to confirm this. But not by a too large margin

[short narwhal]

Veo 3 Fast isn't something you choose in Perplexity; it simply uses it under the hood when you ask it to generate a video

[dry kestrel]

Veo 3 Fast

[silent ocean]

Max users getting screwed over with gpt-5 lately

[short narwhal]

There you go

[dry kestrel]

Thanks for satisfying my adhd brain

[dry kestrel]

They will probably just stick to o3 pro

[agile charm]

Thanks Berd, appreciate it.

[sweet zealot]

that's a shame they don't add gpt-5 thinking ...

[frosty stone]

Manus is too restrictive in credits which would have anyone turned off. You want to feel free and unrestricted

[silent ocean]

If they gave you unrestricted access I'd consider subbing

But for now

And from my tests

It's dissapointing

[frosty stone]

I hope they will, else o3, o3 pro is superior

[dry kestrel]

I doubt any of these "vibe coding" or agent platforms will introduce an unlimited plan

[silent ocean]

I'm confident with perplexity, that if I type a question, for example, I'll get a trustworthy response. With Manus? Not even close.

[blazing comet]

Weekend time

[frosty stone]

True

[sweet zealot]

yeah but whats the point? i'm paying 20 you payin 200 for o3??? like WTH

[dry kestrel]

o3 pro is only available on the $200 plan, same on ChatGPT

[silent ocean]

O3 pro is overshadowed by grok 4 and gpt-5

[sweet zealot]

get my point.

[silent ocean]

Gpt-5 is free for chatgpt users

That's sad

[dry kestrel]

Only the base version - non reasoning one

[silent ocean]

Max users are really getting screwed with

[dry kestrel]

They get 1 free thinking query a day

[silent ocean]

You can trigger reasoning just by adding "think about this longer" at the end of each query

[dry kestrel]

their router was broken at launch - not sure if it's completely fixed now

[frosty stone]

200 to explore the full potential. With pro you really wont re-prompt every lab response until satisfaction

[meager rover]

bro if they replied to you and not my wonderful explanation I give up

[gritty parrot]

sup

[silent ocean]

Sam said some shit about it

Let me link it

[short narwhal]

aahh I missed yours too

[dry kestrel]

Yeah I saw, I need to test it on my free account some more

[meager rover]

holy shit

thats it

pulling out the big guns

[dry kestrel]

https://tenor.com/view/mouxy-cotorep-aveugle-debile-esquiver-gif-20546128

[sweet zealot]

what happened?

[silent ocean]

https://x.com/sama/status/1953893841381273969?t=oV0X7FnJHhdo7PeVdnrXfw&s=19

Sam Altman (@sama)

GPT-5 rollout updates:

*We are going to double GPT-5 rate limits for ChatGPT Plus users as we finish rollout.

*We will let Plus users choose to continue to use 4o. We will watch usage as we think about how long to offer legacy models for.

*GPT-5 will seem smarter starting

X · Aug 8, 2025, 6:59 PM

[meager rover]

Life

[meager rover]

double as in 40 -> 80 or 80 -> 160?

[silent ocean]

We need to add another mode called agent in perplexity

[minor shard]

160

[silent ocean]

Which will do basically what comet assistant does

[meager rover]

ok good

[silent ocean]

But everywhere

[meager rover]

?

[short narwhal]

Why is it thinking for more than 1 message on the free plan for me 🤨

[dry kestrel]

ChatGPT Free tier accounts can send up to 10 messages every 5 hours. After reaching this limit, chats will automatically use the mini version of the model until your limit resets. Free tier users also have access to one GPT-5 Thinking message per day.

[minor shard]

lol the question

[meager rover]

holy shit

[dry kestrel]

You can still get autorouted to the thinking version

[silent ocean]

Yeah that's what I said

[short narwhal]

Am I using mini thinking ehh

[silent ocean]

You can get thinking everytime

[short narwhal]

I don't get it, GPT-5 is strangeee

[meager rover]

that wasnt even like 10 seconds bro oh my god

[silent ocean]

Dissapointing for now

[dry kestrel]

no u i tink matsku strangee

[silent ocean]

I don't see those benefits at all

[dry kestrel]

Also free users get a lower reasoning effort

[silent ocean]

Responds like any other model, nothing special like reasoning models

[short narwhal]

D:

[meager rover]

in perplexity

[meager rover]

guys im a power user

and im shiny shiny

[short narwhal]

my gpt is on a different system

[dry kestrel]

Lmao

[short narwhal]

woaah shinyy

[minor shard]

i love your questions

[prisma apex]

When did uploading videos on ChatGPT get prevented?

[dry kestrel]

It was never a thing

[meager rover]

8k context window its over

[fast meteor]

o3?!

[dry kestrel]

YAP score is not context window

[meager rover]

macos chatgpt app is still not upgraded

[dry kestrel]

hehe

[meager rover]

ok party pooper

[dry kestrel]

They removed o4 mini high 😭

[meager rover]

Okay buddy what the shit

[short narwhal]

Sonnet is a twitter user

[dry kestrel]

Im not too sure how to get the score from GPT-5

[meager rover]

Use a browser

or your phone

[dry kestrel]

I am.......

[dry kestrel]

Tried this as well

[meager rover]

Good luck kiddo

Sam will bless you soon

[dry kestrel]

samsung

[meager rover]

what

that’s my goat

[sweet zealot]

still no gpt-5 thinking.... no respond from PPLX in https://discord.com/channels/1047197230748151888/1047204950763122820

[agile charm]

First result, thanks to your help, using a very short prompt simply asking for animation of the subject and easy camera movement instructions;
https://drive.proton.me/urls/ZDZDY64Q6C#zdRCa1mcgImo

Sorry, Proton automatically included the ad, can't remove it without deleting the url.

Thanks to whomever or whatever removed the ad.

[fast meteor]

can you reproduce this?

i ran 5 queries all came back 65

[dry kestrel]

One sec

I'm getting 30/100

[fast meteor]

but is this reproducible if you start a new chat?

[dry kestrel]

yes

5/5 times

memory, customization, all of that off

[meager rover]

hm..

[odd sentinel]

Can confirm with GPT-5, same answer, static 30 in several ones

[dry kestrel]

You need to use the thinking variant

the non thinking one gives 65

[agile charm]

Frame cap from Perplexity video gen.

[blazing comet]

No comment

[fast meteor]

About 30/100 by default. Want me chattier or terser? Toss me a number. by thinking

[dry kestrel]

Interesting

[blazing comet]

Gpt 5 thinking please show me da score

[indigo valve]

YO I just got Comet to scrape the internet for a beta code for Battlefield 6. This is nuts

[dry kestrel]

It keeps looking at my google drive smh

[fast meteor]

knowledge cutoff june 2024? this cant be right?

[blazing comet]

I just got gpt 5 on the app

[fast meteor]

what are you hiding there?

[blazing comet]

I saw all the modes disappear

[dry kestrel]

berd files obv

[naive sage]

Yeah, not much of the data there actually looks right... from articles:

In terms of knowledge, GPT-5’s main model has a cutoff date of September 30, 2024, meaning it was trained on data up until that point. The smaller GPT-5 Mini and GPT-5 Nano models have a slightly earlier cutoff date of May 30, 2024. While all three can still pull in fresh information via browsing tools (when enabled), the cutoff defines the static knowledge baked into the base model.

So maybe this is GPT5-Mini though?

[short narwhal]

@dry kestrel does kimi have something against me

[dry kestrel]

you killed kimi

[short narwhal]

This sucks

[dry kestrel]

(likely just the dev's vpn disconnected)

[fast meteor]

your bucket is an emi shield

[dry kestrel]

@fast meteor The verbosity is different from the Juice score

[serene sand]

timmy

[dry kestrel]

WHAT

[serene sand]

south park

[dry kestrel]

oh

that thing

[fast meteor]

for sure

[dry kestrel]

What did I just see

hmmm

[fast meteor]

what are your current vibe-o-meter from 0 to 100?

5: 87/100
thinking: 87/100

[short narwhal]

Love vibes

[dry kestrel]

@short narwhal I think I got scammed

[meager rover]

living large

[fast meteor]

is vibe also some kind of hyperparam in chatgpt?

[short narwhal]

You used something that didn't use those free credits OR you went over the limit 🤓

[dry kestrel]

All I used was GPT-5 and chat latest

I blame this on you

[short narwhal]

woahhh relax

[meager rover]

[dry kestrel]

There is no way

I used 700K input tokens

[meager rover]

lol

[dry kestrel]

What the hell

[fast meteor]

huge hidden system prompt?

[dry kestrel]

I don't have anything set

for the sys prompt

[fast meteor]

or is your api secret leaked?

[dry kestrel]

Doubt this is true either - I'll reset it but I don't think it was

I have 2fa on and I never used it

[obtuse lance]

when I go to spaces and chose a thread from yesterday how I know which model was set at that time. I got an excelent answer not even gpt-5 get close to that but I dont remember what model was that, but id like tu use that again.

[dry kestrel]

Oh wait. I think codex used the API instead of my ChatGPT usage

great

[short narwhal]

Can't blame me no more

[dry kestrel]

It still should have used the free credits I get

so its still your fault mat!!

[short narwhal]

:(

[dry kestrel]

thats a bit annoying

[meager rover]

Codex bills to your account usage?

I thought it was only API

[dry kestrel]

It can now

[meager rover]

Unlike Claude code

[meager rover]

Source

[dry kestrel]

internet/github official repo

[short narwhal]

Should be visible in this little processor menu

[meager rover]

[obtuse lance]

I see thx a lot

[blazing comet]

Claude W Custom instructions

[obsidian umbra]

bro

😭

[meager rover]

bro mine routed to thinking just to be careful

[blazing comet]

💀 💀 💀

[empty ice]

Guys

Holy

Use this in cursor

[gleaming plume]

Does anyone know if GPT 5 Thinking will come to Perplexity? I love O3 and it seems to be the direct replacement

[dry kestrel]

Nothing official regarding this so far

[sweet zealot]

they should

[fast meteor]

maybe for ?

[hollow vessel]

just got gpt 5 on my teams account for 1 person (me) for unlimted use

[hot mantle]

https://x.com/sama/status/1953953990372471148?s=19

Sam Altman (@sama)

Wanted to provide more updates on the GPT-5 rollout and changes we are making heading into the weekend.

1. We for sure underestimated how much some of the things that people like in GPT-4o matter to them, even if GPT-5 performs better in most ways.

2. Users have very different

X · Aug 8, 2025, 10:58 PM

[fast meteor]

Wanted to provide more updates on the GPT-5 rollout and changes we are making heading into the weekend.

1. We for sure underestimated how much some of the things that people like in GPT-4o matter to them, even if GPT-5 performs better in most ways.

2. Users have very different opinions on the relative strength of GPT-4o vs GPT-5 (just the chat model, not the advanced reasoning one). This is a cool thing you can try: https://x.com/flowersslop/status/1953908930897158599

3. Long-term, this has reinforced that we really need good ways for different users to customize things (we understand that there isn't one model that works for everyone, and we have been investing in steerability research and launched a research preview of different personalities). For a silly example, some users really, really like emojis, and some never want to see one. Some users really want cold logic and some want warmth and a different kind of emotional intelligence. I am confident we can offer way more customization than we do now while still encouraging healthy use.

4. We are going to focus on finishing the GPT-5 rollout and getting things stable (we are now out to 100% of Pro users, and getting close to 100% of all users) and then we are going to focus on some changes to GPT-5 to make it warmer. Really good per-users customization will take longer.

5. The team is doing heroic work to optimize our systems and find more capacity, but still, we are looking at a severe capacity challenge for next week. We are still deciding what we are going to do, but we will be transparent with our principles. Not everyone will like whatever tradeoffs we end up with, obviously, but at least we will explain how we are making decisions.

Thanks for your patience with us; we will continue to react and improve quickly!

[wise nova]

Team for $25 have same limits as Pro for $200?

[green crystal]

So basically he's saying that plus users should not expect their double rate next week

[dry kestrel]

just normal GPT-5 is unlimited

the non reasoning one

[green crystal]

Oh well, it's pretty decent the current limit

Guess it'll just take longer to get the higher limit

[empty ice]

Use that one in cursor

Crazy good

[wise nova]

But Plus version is limited?

[dry kestrel]

80 /3 hrs

[empty ice]

Another dc users comment:

[empty ice]

Re-made a Thingie i did a while ago with gemini 2.0 pro i believe, which was a Warframe.market Thingie i made for myself to simplify some aspects of trading for warframe, gpt 5-high + cursor made it so much faster and better

[wise nova]

So better to pay for Team than Plus?

[modern crown]

Do you see gpt-5 pro yet? teams plan users should get pro

[dry kestrel]

it would be very very limited

Just get a Pro subscription if you need GPT-5Pro

[modern crown]

I know but you at least get it!

And can do credits useage

So does any Teams users have gpt5-pro yet?

[empty ice]

Guys

I think gpt 5 is like this:

[wise nova]

Does that changes concern API?

[empty ice]

https://tenor.com/view/mamadizimo-gif-26501724

[low talon]

I know this is a shot in the dark but does anyone have a spare comet invite theyd care to share?

[empty ice]

How can u get one?

I have comet

But idk where is the invite

[lethal mulch]

chat this is so infuriating lol i’m actually glad I have perplexity pro instead of chatgpt plus

[empty ice]

Use thinking

It should reroute

[low talon]

thanks for caring!

I know theyre somewhere... ill dm u if youre willing to check?

[empty ice]

Yeah

[sweet zealot]

i'm losing my mind .... everyone like cursor - git copilot and... gets gpt-5 thinking and .....

[empty ice]

Do not rush haha

U have plenty of time

[sweet zealot]

and we got one that's working like no good.

[empty ice]

I mean normal gpt5 earlier eas so bruh level

Now routing is better

Thinking is great ij chat

@low talon check dm

[odd sentinel]

If you have any available, you can find them here https://www.perplexity.ai/browser/invite

[short narwhal]

https://www.perplexity.ai/browser/invite

Yaak I was late

[empty ice]

https://tenor.com/view/cat-gif-4807728093068826302

I found it on yt haha

It worked

Thanks tho

Thank u bois

I have one more if anyone needs

[odd sentinel]

Here, please!

[empty ice]

Oki

Check dm

[serene sand]

Don’t forget to share your card: https://discord.com/channels/1047197230748151888/1394534633328345148

You’ll also receive two invites, it’d be great if you could share them with other community members: https://www.perplexity.ai/browser/invite

🙏

[empty ice]

Cards look so cool

[languid star]

SimpleBench

[wise nova]

I like Gemini in pplx

[meager copper]

The best

[wise nova]

Grok4 not working:
Used Grok 3 Beta because Grok 4 was inapplicable or unavailable.

[quaint ravine]

i think it's still the best all around especially combined with Perplexity's RAG

although perspectives can be heavily biased to what your use cases are

[empty ice]

Eh bs for me

Idk what the used

[wise nova]

Wtf, how many times I get Grok3 answers instead Grok4 and not be noticed

[empty ice]

But grok 4 and gpt 5 high is like fat man vs tsar bomb

Idk why people glaze gemini that much besides search

It is literally worse to code

Search is great yeah

Creative writing is truly bad on gemini

Grok claude gpts all better

Even mistral

[green crystal]

It's the worst at search

[empty ice]

Huh

[green crystal]

Even pplx is better

[empty ice]

I mean pplx+gemini 2.5 pro

[green crystal]

Ah

[empty ice]

Pplx is p good haha

Gpt 5 is not bad in pplx

But just that

Not great either

But not bad

[green crystal]

It's limited and non reasoning in pplx

[empty ice]

Gemini AI tried writing a novel.
Chapter 1: "Once upon a time…"
Chapter 2: "Refer to Chapter 1 for similar content."

[quaint ravine]

btw GPT4.1 still accessible in Perplexity depending on your prompt 😛

[empty ice]

[cosmic forum]

Anyone else got video generation on perplexity??

[sweet zealot]

i just used gpt-5 on git copilot... damn....

@vague pagoda please add the fat gpt-5 model in pplx for pro users .... the gpt-5 pro should be for the max users too :((

[cosmic forum]

GPT-5 is already on perplexity

[sweet zealot]

not the thinking model...

https://discord.com/channels/1047197230748151888/1403076559207796927

https://discord.com/channels/1047197230748151888/1403268394475716679

[empty ice]

Also

[naive sage]

Its explanation isn't completely right (it says subtracting 5.11 from both sides after it's already done it by isolating x), but at least it gives the right answer.

[empty ice]

Does anyone have a student status

[empty ice]

I can give u my 1 month referral

[naive sage]

So what do you think causes GPT5 to get it wrong sometimes, is it all just randomness.

[languid star]

hallucinations of model

is a fact

its low in gpt 5

[naive sage]

Maybe eventually models will get fast enough that they can run n-times in parallel and then check themselves for more accurate responses.

[languid star]

like grok 4 heavy

[naive sage]

Yeah, I guess so... kinda like that.

[quaint ravine]

presumably this applies to Perplexity? "Developers can control GPT‑5’s thinking time via the reasoning_effort parameter in the API. In addition to the prior values—low, medium (default), and high—GPT‑5 also supports minimal, which minimizes GPT‑5’s reasoning to return an answer quickly."

or the current model has 0 reasoning? hence why it's so fast I guess

it's actually quite impressive how fast it is for the answers it gives

[hollow vessel]

No

[opal osprey]

Can I ask here if everyone gets the quirk with Comet being unable to clear inputs or is it only Windows users?

[modern crown]

It will show up eventually

[opal osprey]

It's kind of a major issue that gets in the way of nearly any task you give Comet. I just want to know if it's a Windows thing or is it everyone?

@minor shard, don't mean to ping you, but just curious if you get this as a Mac user with Comet

[empty ice]

Comet is so 50 50

Sometimes it is scary gpod

Sometimes it is so goofy i have to laugh

[opal osprey]

lol, most of the time I find it very good EXCEPT this input issue

@empty ice, do you get this input issue?

[empty ice]

What is that

I find the pdf summaries lacking tbh

They are way too short

I am not a lawyer but i use comet to fund laws often

It usually can

[opal osprey]

if you go here: https://comet-input-tests.netlify.app/
tell the Assistant: "Can you do Test 2?"

[empty ice]

https://tenor.com/view/what-what-the-what-the-hell-wtf-gif-8562994133044611418

[dapper zealot]

anbird on my head

[opal osprey]

it's just a test thing to see if Comet can do inputs properly

[empty ice]

I see haha

[opal osprey]

does it work for you?

[empty ice]

I am on phone 🙁

I will try tomorrow

[opal osprey]

oh dont worry then

[empty ice]

[empty ice]

It can input for me tho

[opal osprey]

yea i didnt even set the site up for phone

[empty ice]

But at first

It tries to fill in my prompt

I to the bars

Then deletes it

[opal osprey]

ah okay yea

[short narwhal]

Seen a few people have the same issue

[opal osprey]

do you have it? i'm guessing not if you haven't noticed it?

because if you get it, it pretty much happens on every single input 😆

[short narwhal]

I feel like I’ve had it sometimes struggle deleting something? It had a hard time in Google Docs for me some time back

[opal osprey]

yea that is the issue, though at least for me, it happens every single time

well very, very occasionally it actually does do that part right, but it's something like 98% of the time it wont

i've been testing Comet on all my standard things, and if i help it with the input parts it pretty much does everything else right

[tardy sparrow]

Hi, is there a way to get access to comet without paying the $200 / month ? I am on a pro plan.

[dry kestrel]

#💬│general message

[frosty stone]

Yes but limited agentic access

[tardy sparrow]

How can i get the limited version ?

[frosty stone]

Get an invite or wait

[tardy sparrow]

Would you happen to have an invite?

[twin grail]

what do you mean ?

Comet ?

If that's the case you can ask for an invitation at #☄️│comet-general

[frosty stone]

Yes but its nothing extraordinary given youd have to pay for the functionality

[twin grail]

Or you can install Comet and use it but the agentic feature will be limited

[tardy sparrow]

Thank you.

[tardy sparrow]

Could you kindly share the invite ? I have been using Dia. I have been curious to check it out.

[lusty swift]

is GPT performing better since new update from Altman?

[sweet zealot]

24-hour mark passed, no GPT-5 thinking yet... I tried it both in Copilot agent mode and in OpenAI itself. It's amazing, but the only paid plan I have is PPLX. That's bad because they say they are the fastest, but when it comes to something as important as GPT-5, they lack providing us with information after 24 hours of its official release.

[languid star]

https://www.youtube.com/watch?v=_EonL-sReug&ab_channel=ENFANTTERRIBLE

can i alone make a game like this ?

[lusty swift]

?

[lusty swift]

everything is possible

the limiting factor is the time

[tardy sparrow]

How can i do that ?

[twin grail]

U can

But the thing is

are u willing to take it ?

U can literally clone anything

u just need time

[lusty swift]

Bro you’re corny

😭

[dry kestrel]

lots of time and resources

[twin grail]

Ye kinda

[languid star]

https://tenor.com/view/disintegrating-aughhh-gif-24532719

[twin grail]

also sanity

[lusty swift]

Is GPT performing better since new update?

[twin grail]

i can share you the installer if you want

[dry kestrel]

at least in ChatGPT yes

[lament lark]

Bro how y'all online all the time💔

[lusty swift]

You mfs answer

[dry kestrel]

Chill

[twin grail]

in GPT yes

[foggy valley]

Ya GPT5 is the shit

[lusty swift]

I mean in pplx

[lusty swift]

What you gon do

[dry kestrel]

What do you want me to do?

[dry kestrel]

Haven't noticed any significant improvements/changes

[lusty swift]

I see

[twin grail]

relax buddy

[lusty swift]

You gonna relax

Lil bro

[twin grail]

timing

[sweet zealot]

We should all leave message here so they add this correctly and quickly
Other added the thinking mode asap
PPLX not and just add a crap version of it for us so if we let this slip, they will forget. Like always....

https://discord.com/channels/1047197230748151888/1403076559207796927

[twin grail]

the API cost is insane

[lament lark]

Scam Altman strikes again

[twin grail]

even for Plus itself

paying for 20$ a month still having limit

[dry kestrel]

?

[twin grail]

Only team version having unlimited access

[sweet zealot]

https://discord.com/channels/1047197230748151888/1403268394475716679

[foggy valley]

see my video

[twin grail]

So for 20$ a month u can't ask for 600+ ask per day

for a new thinking model

[dry kestrel]

It answered the question correctly?

[lament lark]

Yes

[opal osprey]

Slightly offtopic, but in OpenRouter, to use the reasoning GPT-5 you need to add an OpenAI key. When I do this, I still seem to get an error when trying to use it

[twin grail]

please read.

[dry kestrel]

why are you using openrouter for OAI models in the first place

[twin grail]

Even OpenAI itself have limits on their new thinking model

I have plus

[languid star]

sometime it says 2 and sometimes 3

[foggy valley]

🙁

[twin grail]

Both plus and team plan

[twin grail]

20$ and u still get being limited with GPT 5 Thinking

So what do u think for Perplexity adding it with 600+ ask per day

[foggy valley]

how many r's are in RIP

[twin grail]

for only 20$ and other models ?

Nah i don't think that's possible

It's more about the cost and how they gonna make profits

[lusty swift]

Sonnet thinking answers correctly

On this question

[twin grail]

it's a reasoning model

ofc it will

[lusty swift]

ChatGPT answers wrong

[lusty swift]

And GPT 5 is not reasoning model lil bro

Watch your mouth

[foggy valley]

😮

[opal osprey]

hmm i was using it for all my models? i have my own interface i had made to access them through the browser

if i was subscribed to OpenAI Plus i'd probably use it on the OpenAI platform. I get what you're getting at you get benefits like how the canvas and everything is integrated. But i also wanted to be able to have a centralised place where i can do everything

[dry kestrel]

the API is very different to the subscriptions on ChatGPT

[opal osprey]

yea i know what you mean, i still wanted to get it set up working though

[dry kestrel]

There is a reasoning and non reasoning variant.

[sweet zealot]

yeah ik about all of this but no matter what they added a crappy version of the API for the PPLX. everyone talks about it like 90% of the chat is about the gpt-5 and thinking models and how bad it is rn in pplx

[dry kestrel]

You still need an OAI account with funds in it to be able to use the model via openrouter

[twin grail]

it's not crappy

Perplexity was meant for " Researching "

So on their end they add system prompt

[opal osprey]

maybe that is why its rejecting, that is probably it. In that case i might add some. I thought i had some in there... perhaps not

[dry kestrel]

You need funds in your OAI account for it to work.

[sweet zealot]

You mean it works okay in space with proper rules?

[twin grail]

and how they would actually perform

[opal osprey]

haha actually i do have funds in there already, damn

i'll keep trying

[dry kestrel]

oh thats weird

hang on

[twin grail]

So it might change the way how it answer you

[dry kestrel]

I'll try it myself

[twin grail]

yes if you can prompt it correctly

[sweet zealot]

so they don't nerf the API from there side just tweak it and we can make tweak for our self???

[opal osprey]

wow, thanks, appreciate that big time

[twin grail]

The reasoning model itself isn't crappy

it's how you use it and what's u expect from it

[twin grail]

It's not a tweak, it's more like a system rule that the AI gonna follow

And perplexity was build more for researching

[sweet zealot]

so the API is the same?

[twin grail]

So when it come to solving other tasks it might not perform well

Yes it's the same

[dry kestrel]

I was able to get GPT-5 Chat to work, not the reasoning models - my guess would be you would need to verify your organization on OAI

[sweet zealot]

Bro... you can cook anything in the oven.... - as long as it has Sonnet 4 and... it's good for coding too.

i tought they just limit the API and nerf it for there cost....

[dry kestrel]

Yes - they can change things like temperature though which could affect outputs

[opal osprey]

yea i've got the GPT-5 Chat, Mini and Nano working currently, but not the reasoning as you said. I appreciate you testing this! I will look at the organisation thing now. I'm getting deja vu, I think I looked this up before with o3 pro

[dry kestrel]

Np

[neat glacier]

I knew it GPT 5 is really bad https://www.reddit.com/r/ChatGPT/comments/1mkae1l/gpt5_ama_with_openais_sam_altman_and_some_of_the/

[dry kestrel]

Uhhh no? Most people are just mad that their ability to choose themselves what model to use was taken away

[opal osprey]

Ah yea, it's definitely the organization verification causing the issue, the error message contains this:

"Your organization must be verified to stream this model. Please go to: https://platform.openai.com/settings/organization/general and click on Verify Organization. If you just verified, it can take up to 15 minutes for access to propagate"

I might do that later, it seems it doesn't matter if you're an individual / organization to do this

[dry kestrel]

Yeah, they just want your id :/

[wise nova]

How to check if the Gemini 2.5 Pro in pplx are thinking model?

[wary idol]

Hello friends, does anyone have an invitation to the comet? If not, sorry for the inconvenience, I wish you all a great night

[dry kestrel]

You can tell since it doesn't give you a response right away

[sweet zealot]

maybe they add waiting point for it from the back-end side so it responds longer, making you think it's thinking? :((

[dry kestrel]

That would make no sense and you would be able to tell that it's not 2.5 pro given the response quality

[wise nova]

Showing status Resesrching..

[sweet zealot]

i both have Gemini Pro Plan and PPLX

tell me what should i search or....

and i'll give you the output answer.

[native sluice]

uh. where is gemini 2.5 pro gone? i was using till today now its not there with others thinking models anymore...

[sweet zealot]

@dry kestrel @wise nova

[dry kestrel]

???

[sweet zealot]

i have both PPLX and Gemini Pro Plan
give me a prompt that i send to both of them to get output.

[sweet zealot]

it's here

[dry kestrel]

I'd try asking Perplexity for any prompts or looking up on reddit.

[hallow tulip]

Does the GPT-5 on Perplexity use thinking or are they using the base model GPT-5 (non thinking) want to subscribe not sure though

[wise nova]

Ask to analyze and merge this two prompts and create a super prompt

[sweet zealot]

https://discord.com/channels/1047197230748151888/1403076559207796927

😄

[sweet zealot]

PPLX finished....

the Gemini still thinking.

even now.

https://g.co/gemini/share/a894274ecac8

https://www.perplexity.ai/search/analyze-and-merge-this-two-pro-jhX_.0.aTTG.RzGEdPdFnA

[wise nova]

Does it Free or Paid version?

[sweet zealot]

paid

[native sluice]

but it was in reasoning section. and it was gemini 2.5 pro... well i suppose 2.5 pro is still this then because i think pro doesn't exist as dumb model

[quaint ravine]

for my use cases, I think it's much better than GPT4.1 (which is what we had before for a while in Perplexity)

[sweet zealot]

Damn

[gleaming plume]

Is this limit of being able to make only 2 videos per month as a pro user temporary? I found it extremely low given Perplexity's history of being generous.

[sweet zealot]

haven't hit it since the last 2 weeks.... what have you done PPLX 😂 you made me switch back to gemini .....

[dry kestrel]

They are using the base model

[dry kestrel]

It's 3 a month

[quaint ravine]

yea the strawberry question on ChatGPT platform actually does a bit of thinking

the Perplexity one is non-thinking, but still surprising it gets it wrong lol

this on ChatGPT platform

[green crystal]

soft limited

prompting can get you thinking most of the time unless its incredibly dead simple

[modern storm]

after many prompts since yesterday, gpt-5 keep destroying all competitors 😅

King of AI 👌

[solar flare]

Hello

[severe sequoia]

Hi

[quaint ravine]

by default, ChatGPT platform gets it right though, as it employs thinking - but if you click "Get a quick answer" fast enough, that's the result....GPT5 "trained" intelligence i guess

[maiden shell]

Hi

[slender berry]

why did perplexity cheap out and put the non thinking gpt5??

[quaint ravine]

it's essentially the replacement of the previous GPT 4.1?

seems reasonable to me

[minor shard]

4.1 wasn't a reasoning model

[quaint ravine]

exactly

[slender berry]

should have added gpt5 reasoning though

[twin grail]

If they add GPT5 thinking

It would only be for max

As the cost for GPT5 thinking model

is way surpass o3

in term of outcome

Even for Plus user having extend access

and team user having flexible access to it

Do u think for 20$

perplexity also have to handle other models

[young hollow]

Will Perplexity Add GPT-5 Pro or Reason?

[twin grail]

providing them all at once

adding GPT5 Thinking won't be a great choice in term of making profits

they will be actually losing money

[twin grail]

wdym ?

[young hollow]

Shouldn't GPT-5 Pro Be Max

[young hollow]

As I Said Bro

[twin grail]

GPT5 Pro is only for their exclusive plan GPT Pro for 200$ a month

Do u think OpenAI gonna leave perplexity adding it

to their max plan for unlimited use

The answer is right there

[young hollow]

Ahhh Smart

[twin grail]

they won't be making any profits adding GPT5 Pro with unlimited access

[young hollow]

So We Just Have GPT-5 for Now

[young hollow]

Until The Release of O4

[twin grail]

Even OpenAI itself

have limit on gpt5 pro

for pro users

[young hollow]

I See

[slender berry]

this isnt true

[young hollow]

So Perplexity Pro Users Have Base GPT-5 Pro for Now Until The Release of O4

[slender berry]

o4 isnt even coming out

pretty sure it isnt

[young hollow]

It Should Sometime in The Future

That's What Their Patterns Say

[slender berry]

they said they want to keep all their models unified

[dry kestrel]

The model in PPLX is GPT-5 Chat. This is the official naming from OAI

[slender berry]

not single anymore

[young hollow]

Sorry I Had Meant 5 I Typed Pro By Mistake

[young hollow]

Honest Question

Is GPT-5 Enough for Pro Users?

[slender berry]

my bad forgot they lowered prices, but they hosted o3 to pro users before the price drop lol

[twin grail]

we still got o3

o3-pro for max users

[slender berry]

i know, but gpt5 reasoning is better than o3

[twin grail]

But the price ?

They gonna need to make profits

[young hollow]

How Does 03 Compete to GPT-5 Thinking?

[slender berry]

??? what are u talking about man, they literally host sonnet 4.0 with thinking for pro users

[twin grail]

for 20$ a month with claude, gemini, chatgpt

[slender berry]

and its more expensive than gpt5

[twin grail]

they can't afford to have another reasoning model for 20$ a month

[young hollow]

We Have Both Sonnets They Might Have Wholesale Discount From Anthropic

[twin grail]

Then it's the choice of having claude 4.0 sonnet thinking or gpt5 thinking

[young hollow]

Then Can Remove O3

[twin grail]

U gotta think about how they's gonna make profits

[slender berry]

bro 😭

[twin grail]

providing people with 600 pro ask a day

[young hollow]

And 500 Research

[twin grail]

Even OpenAI itself have limits on their 20$ plan

[slender berry]

do you think adding another model will increase costs? it would literally decrease costs if less users use claude

[young hollow]

Don't Forget Airtel Colab

[twin grail]

this also

so think about it

before spamming crazy about gpt5 thinking

[young hollow]

All The Airtel Users are Currently Using it for Free Like When Jio Had Given Free Data

[slender berry]

?

[twin grail]

for 20$ i'd say this is a good deal

[young hollow]

More then Enough for Individuals

[twin grail]

It's a non thinking model ?

[slender berry]

medium is medium reasoning effort lil bro

😂

[young hollow]

How About They Remove GPT O3 for GPT-5 Reasoning?

[lavish needle]

bro no way i only get 32k tokens of context on plus

thats so sad

[young hollow]

Use File Upload for Refresh Context

[twin grail]

200 a week for gpt 5 thinking

perplexity provide u with 600 pro search a day

how's they gonna make profits

with other models

DR

Lab

[slender berry]

genuinely 0 iq i cant bruh

[lavish needle]

i agree with u SO MUCH

its so stupid

[young hollow]

Maybe They Can Remove Both 03 and Sonnet 4 Thinking Since is Very Similar to Base Sonnet Without Major Improvements and Give GPT-5 Thinking Limited

[gilded finch]

Dleep

Sleep yall

[twin grail]

they won't

because the Claude 4.0 sonnet thinking

is what tempting user to buy

We use perplexity because it provide various models

[young hollow]

I had Seen it has Only 1% Improvement in Coding

[young hollow]

Again Perhaps More Models are Better and Cheaper

[twin grail]

So now if u removed Claude 4.0 sonnet thinking just for new gpt 5 thinking ?

[slender berry]

or maybe they could just remove grok 4 😂

[twin grail]

Nah

It will only be grok 4 and gpt

they will lose customer

[young hollow]

No Grok 4 has Exponential Real Time Data Ability and Only Ai with X Data Acess

[young hollow]

I Think They Would Also Need to Remove O3 to Be Profitable

[twin grail]

Again, if they add a new model they would have to removed the old model

[gilded finch]

Does GPT 5 in Perplexity works the same as Orginal GpT app?

Jus curious

[slender berry]

im trolling

[gilded finch]

[twin grail]

But yes again adding gpt5 thinking won't be possible because of the limits

and the price

[slender berry]

theres no limit

[young hollow]

But We are Serious

[slender berry]

and the price is less expensive dude

[quaint ravine]

unless OpenAI make a special deal with Perplexity

[young hollow]

Best Option Seems to be Removing GPT O3 and Sonnet Thinking

[twin grail]

I don't see why people asking 20$ for a bunch of models

[slender berry]

maybe because gpt5 is literally the best model out now? lmfao

[twin grail]

It's ur opinion but not the rest

they love claude 4.0 sonnet

[slender berry]

o3 should be replaced with gpt5 reasoning

[young hollow]

Tbh

[slender berry]

thats my take

[twin grail]

then u should buy chatgpt plus then

[slender berry]

idgaf about their profits

they have enough vc money 😂

[twin grail]

suit urself then

[young hollow]

The Demand of Profitable Users Will Drive The Future Changes Anyways

The Think is

[slender berry]

i like having other models

im not tryna act entitled

[young hollow]

Around 360M Airtel Users are Now Using Perplexity Pro for Free

For 1 Entire Year

[slender berry]

but still, even adding gpt5 reasoning wont even hurt them

[young hollow]

Wrong

[slender berry]

they could literally add gpt5 (low) and i would still be good

[slender berry]

??? do you think ALL 360m airtel users actively use perplexity

[young hollow]

We are Not Business Analyst to Give Valid Figures

[young hollow]

Honestly Hardly 36 Million

[slender berry]

obviously but im just giving a common sense take

they've added more expensive models in the past

[young hollow]

Wouldn't GPT-5 Reasoning Low Just Be O3 Essentially?

[slender berry]

no its still different

in its speaking, code, etc

[young hollow]

Tbh Seeing Their Patterns

[dry kestrel]

Not exactly a direct comparison

[young hollow]

I Think Eventually They Will Replace O3 with GPT-5 Reasoning

[slender berry]

hope they do

[young hollow]

Well Demand Drives Changes

Most Users Will Demand GPT-5 Reasoning

[slender berry]

perplexity faq shows gpt as reasoning when they have the version that DOESNT reason

[young hollow]

I Think GPT-5 has Reasoning

But is Not as Powerful GPT-5 Reasoning

[slender berry]

no it doesnt

the plain version doesnt

[young hollow]

@dry kestrel Help

[young hollow]

Maybe They Have A Tuned Model Like R1 1776

But if That was The Case The Would be Different

[slender berry]

no they dont

what are u talking about

theres no fine tuned gpt5

well gpt-5-chat but thats by openai , perplexity can't finetune it

[dry kestrel]

?

[young hollow]

@dry kestrel This

Why Does it State Reasoning?

[dry kestrel]

It says expert level reasoning, not that its a reasoning model

[young hollow]

So it Does has Reasoning

[dry kestrel]

No

[slender berry]

"unified system that automatically decides when to use deep reasoning or quick replies"

[dry kestrel]

It uses the non reasoning variant

[slender berry]

router isnt even available since they use gpt-5-chat

[young hollow]

Confusion and Contradictions

[slender berry]

unified system = router that they dont have

[dry kestrel]

One sec

This might be a typo/bug I'm not too sure

[languid star]

[slender berry]

they just asked perplexity "give a description for gpt5" and thats it

lol

[dry kestrel]

I did the best I can, I'm sure that they will look into the wording of that

[muted jackal]

I am already using comet but still got "Try Comet before your invite expires" should I ignore it?

[slender berry]

bro called corporate

[dry kestrel]

https://tenor.com/view/among-us-sus-gif-2558655191726498734

[muted jackal]

.

[languid star]

@dry kestrel gpt5 or claude reasoning

[muted jackal]

GPT 5 ig better bro

[dry kestrel]

Depends on what platform and what you are trying to do

[muted jackal]

@dry kestrel

[dry kestrel]

GPT5 with reasoning has been really impressive for me

[muted jackal]

agree completely

[dry kestrel]

I'm sorry but I'm really not too sure, if you already have access I'd assume you are good.

[muted jackal]

did you ever got it?

[dry kestrel]

No

[muted jackal]

im guessing how powerful will GPT 5 reasoning would be

[dry kestrel]

So far at least it's been really good at frontend stuff

[muted jackal]

claude was not good even before if you see

[sweet zealot]

i used sonnet 4 and gpt5 in git copilot. gpt-5 is like 3X faster and fix the issue in one shot !!!!

[young hollow]

I don't think Base Sonnet Can Compete

Only Opus Can Compete with GPT5

Mainly Opus 4.1

@dry kestrel Why I Can't Select Model for Research?

[dry kestrel]

It was never supported for Deep research.

[young hollow]

Any Workaround?

[dry kestrel]

Not really, you can only switch models using the normal pro search.

[sweet zealot]

@dry kestrel Can I know the difference between GPT-5 in OpenAI itself and the one we have? I want to use it for my tasks, but it has a lot of issues right now, even with correct spacing.

[young hollow]

Gotcha I Can Use Pro Search for Usage of Each Model and then Do Research With All That Data

[dry kestrel]

The GPT-5 model you get on PPLX is the exact same base one you get on ChatGPT (the one in the screenshot). The difference is that on ChatGPT, it can determine if the query is difficult enough and if it is, they will automatically route you to GPT-5 thinking.

[sweet zealot]

ok but dosent the thinking model the one with the most hype and...? like how much does the gpt-5 that we have/they have in normal is better than other models in PPLX? like Sonnet 4 thinking or gpt - 5 now?

[dry kestrel]

I'd stick to sonnet 4 for now on Perplexity and use GPT-5 for quick answers

[young hollow]

Well At Least When GPT-6/Opus 5 Release Previous Models Will be Available for Pro Users

[sweet zealot]

:)))))))

[young hollow]

Is Just A Game of Time and Patience

[young hollow]

It Has Always Been The Industrial Pattern

[twin grail]

__~1

[young hollow]

Enterprise > Max > Pro > Free

[twin grail]

it's actually Max > Enterprise > Pro > Free

[young hollow]

Sorry

I Thought Since Enterprise Pay More They Should be First Priority

[twin grail]

For max, 200$ a month for personal use

for enterprise depends on the slot

5-50

For 5 like mine

[young hollow]

Teams Can Use it

Since is Literally Unlimited

[twin grail]

It's 2000$ a year similar to enterprise

[twin grail]

There's no Enterprise Max

Only Enterprise Pro

[young hollow]

I Know Im Just Saying Small Teams Cam Use Max

Since is Unlimited

[twin grail]

Yes and company is using it

but the think when people buy enterprise

[young hollow]

1 Account for All Team Members

[twin grail]

is other feature for enterprise

which is not on Personal use

[young hollow]

Like API Acess?

[twin grail]

more than that